Generative AI Hallucinations in Legal Briefs
Courts are sanctioning lawyers because generative AI tools fabricate case citations, quotations, and holdings that do not exist, and attorneys are filing them without verification. Judges view this as a failure of the duty of candor owed to the tribunal. When a brief cites a nonexistent decision, opposing counsel and the court waste time chasing phantom authority, and the integrity of the adversarial process suffers. Sanctions range from fines and mandatory legal education to referrals for discipline, as seen in Connecticut and Ohio matters.
Also worth reading: How does AI legal document verification work and what are the risks of hallucinated citations in court filings? · How Should Lawyers Review AI-Generated Legal Citations in 2026? · Can courts sanction a party for AI-related spoliation in eDiscovery, and what do lawyers actually need to do in 2026?
The problem is not the technology alone but the absence of professional diligence. Lawyers remain responsible for every citation they sign, and several courts have warned that “the AI made me do it” is no defense. Some judges now require certification that no generative AI was used in drafting or that all citations were independently verified. The recurring lesson from bar associations and court opinions is straightforward: treat AI output as a lead, never as authority, and confirm every source before it reaches a filing.
Court Sanctions Across New York and Ohio
Courts in New York and Ohio are sanctioning lawyers who submitted briefs containing AI-generated citations to cases that do not exist. Generative AI tools can produce convincing but entirely fabricated authorities, a phenomenon known as hallucination, and attorneys who file these fabricated citations without verification face consequences ranging from fines to mandatory legal education. The New York State Bar Association has warned practitioners to look beyond the mirage of generative AI, emphasizing that polished language and confident formatting do not make a citation real. In Connecticut, a court sanctioned a lawyer for hallucinated citations and required both continuing legal education and a charitable donation, a outcome the Hartford Courant described as a reminder to the entire profession rather than a punishment aimed at one attorney.
In Ohio, two attorneys faced sanctions after filing briefs containing AI-generated falsehoods, including invented case law and misquoted decisions. ABA Journal reporting shows a clear trend: sanctions grow harsher when lawyers lack candor about how the errors occurred, attempting to blame staff or hide their use of AI rather than admitting it. Judges increasingly treat honesty about AI use as an ethical obligation. For firms using AI eDiscovery, legal research, and document drafting tools, the lesson is straightforward: AI can accelerate work, but every citation, quote, and holding must be verified by a licensed attorney before filing.
Lack of Candor Leads to Stricter Penalties
Courts across the country are sanctioning lawyers who submitted briefs containing fabricated case citations generated by artificial intelligence tools. In Connecticut, a judge sanctioned an attorney for filing a brief with hallucinated citations, ordering him to complete legal education courses and make a charitable donation—a penalty the Hartford Courant described as a "reminder" to the entire profession rather than punishment of one lawyer. Similarly, two Ohio attorneys face sanctions after filing legal briefs containing AI-generated falsehoods, demonstrating that these incidents are not isolated events but part of a troubling pattern.
What has drawn judicial ire is not merely the use of AI itself, but the lack of candor about it. As the ABA Journal reports, attorneys who initially blamed AI errors or failed to disclose their reliance on generative tools have received stricter sanctions than those who admitted mistakes promptly. The New York State Bar Association's warning to "beware of generative AI and hallucinations" captures the core lesson: lawyers remain fully responsible for verifying every citation and assertion. As courts embrace AI's potential for legal research and drafting, they are making clear that efficiency never excuses diligence, and honesty about technology use is now a professional obligation.
AI eDiscovery and Legal Research Risks
Courts across the country are sanctioning lawyers who filed briefs containing fabricated case citations produced by generative AI tools, and the penalties are escalating. In Connecticut, a judge sanctioned an attorney who submitted a brief with hallucinated citations, ordering legal education courses and a charitable donation—a punishment the Hartford Courant described as a "reminder" to the entire profession rather than a one-off punishment. In Ohio, two attorneys face sanctions for legal briefs containing AI-generated falsehoods, showing that state courts are actively policing this conduct. The New York State Bar Association has warned practitioners to beware of generative AI and hallucinations, while New York's judiciary itself is exploring AI and the courts as new frontiers in legal writing and research for judges.
What troubles courts most is not the initial error but the cover-up. The ABA Journal reports that attorneys' lack of candor over AI errors leads to stricter sanctions, because lawyers who double down after being caught mislead the tribunal. The lesson is clear: verify every citation, disclose AI use when required, and treat AI output as a draft, never an authority.
Best Practices for Safe AI Document Drafting
Courts are sanctioning lawyers over AI-assisted brief writing because generative tools can fabricate case citations, quotations, and holdings that appear authoritative but do not exist. In one widely reported matter, a Connecticut attorney faced sanctions and was ordered to complete legal education and make a charitable donation after submitting a brief containing fictitious authority. Similar proceedings in Ohio and elsewhere have underscored that the problem is not isolated to any single jurisdiction or practice setting.
The deeper issue is candor. When lawyers fail to verify AI output and then decline to correct the record, courts treat the conduct as a breach of duties owed under professional rules, prompting stricter penalties. Judges increasingly expect attorneys to disclose how AI was used, confirm that cited sources are real, and stand behind every representation. Tools that support AI eDiscovery, legal research, and document drafting must therefore be paired with rigorous human review, because the sanction risk falls on the signing attorney, not the software.
AI Legal Misconduct Cases and Sanctions Compared
| Case / Jurisdiction | Misconduct | Sanction / Outcome |
|---|---|---|
| Connecticut lawyer (Hartford Courant) | Filed brief with AI-hallucinated citations; called a "reminder" to the profession | Court sanctions, mandatory legal education course, and charitable donation |
| Two Ohio attorneys (NBC4 WCMH-TV) | Submitted legal briefs containing AI-generated falsehoods and fake case law | Facing formal sanctions and disciplinary proceedings |
| Attorneys cited by ABA Journal | Failed candor when questioned about AI errors in filings | Stricter sanctions imposed due to lack of honesty, not just the errors |
| New York (NYSBA guidance) | "Beyond the Mirage" report warns of generative AI hallucinations | Professional guidance urging verification duties for all practitioners |